But it does not exist in real life. It is a face generated on a website – aptly titled that person does not exist – by artificial intelligence. If you reload the page, it will be replaced by another face that is equally attractive, but equally unreal.
Released earlier this month by software engineer Phillip Wang as a personal project, the site makes use of an artificial intelligence system recently released by researchers at Nvidia. Named StyleGAN, the AI is adept at creating some of the most realistic faces of non-existent people the machines have produced so far.
This person does not exist is one of several websites that have appeared in recent weeks using StyleGAN to produce images of people, cats, anime characters and vacation homes that seem closer and closer to reality and, in some cases, are indiscernible for the common spectator. These sites show how easy it is for people to create fake images that look plausibly real – for better or for worse.
Wang, like many researchers and AI enthusiasts, is fascinated by the potential of this type of AI. So much so that he created a second site called thiscat non-existent that spawns fake felines. But He is also worried about how this could be misused.
This makes sense, since the AI tactic underlying StyleGAN was also used to create so-called "deepfakes", which are persuasive (but false) video and audio files intended to show a real person doing or saying something they did not do .
These concerns are echoed by prominent voices in the industry. Earlier this month, research firm for nonprofit OpenAI AI decided not to launch an artificial intelligence system created by it, citing fears that it is so good at composing text that could be misused.
But even if the images appearing on Wang's website can be used to, say, help a scammer create realistic online characters, he hopes that it will make people more aware of the emerging capabilities of AI.
"I think those who do not know about technology are more vulnerable," he said. "It's like phishing – if you do not know, you can fall for it."
The fascination (and counting) of false people
Many people are not sure how to feel about such easy access to fake faces. But they are interested in seeing them.
Wang, who was previously a software engineer at Uber, was studying AI on his own for six months when he set up his website in February – shortly after Nvidia publicly released StyleGAN. He posted on the site in an AI group on Facebook on February 11. In the following weeks, about 8 million people visited him.
"I think for a lot of people out there, they look at it and say," Wow, Matrix! "This is a simulation?" People really are on the computer? "Wang said.
The generator creates a new face every two seconds, Wang said, which you'll see when you refresh the page.
"You can think of it as the AI is dreaming of a new face every two seconds on the server and displaying it to the world," he said.
The faces that visitors see vary infinitely, with a multitude of eye colors, face shapes and skin tones. Some wear lipstick or shade; a handful of sporting glasses. Occasionally, a guy with facial hair appears; one even looked sweaty.
They have all kinds of facial expressions. Some smile, others pout or look serious. The younger faces appear to be children, but none appear to be older than middle-aged.
As realist how these faces may appear, there are still many details that reveal that they are not real people. For example, teeth often look a bit strange and as if they are in dire need of handsets, and accessories like earrings could appear in just one ear. Often, a person seems to have a supernatural skin condition or serious facial scars. The clothes may seem blurry, have swirls of colors, or kind of, odd.
How faces are made
In order to generate these images, StyleGAN makes use of a machine learning method known as GAN, or generator network adversary. GANs consist of two neural networks – which are algorithms modeled on neurons in a brain – facing each other to produce real-looking images of everything from human faces to Impressionist paintings. One of the neural networks generates images (of, say, a woman's face), while the other tries to determine whether this image is false or real.
Although the field of AI goes through decades, the GANs have only existed since 2014, when the tactic was invented by Google's research scientist, Ian Goodfellow. They quickly gained prominence among many researchers as a major breakthrough in the field.
StyleGAN is particularly good at identifying different characteristics within images – such as hair, eyes and face shape – which allows people to use it to have more control over the faces that appear. This can result in better-looking images as well.
The pretense produced by the GANs can be fun – if you know what you're looking for – and potentially big companies. A startup named Tangent, for example, says it is using GANs to modify faces from real-life models so online retailers can quickly (and realistically) customize catalog images for buyers in different countries, instead of using different models or Photoshop . A video game company can use the GANs to help create new characters or interact with existing ones.
This is not an Airbnb
Christopher Schmidt, a software engineer at Google, was one of millions of people who saw Wang's site shortly after launch. He noted that the Nvidia researchers had also trained StyleGAN to create realistic bedroom images and had the idea of building their own non-ritual site to combine AI counterfeit images with AI-generated text. The text generator he used was trained on several Airbnb lists.
Nvidia declined to comment on this story. A spokesman said that this is because the company's StyleGAN research is undergoing peer review.
Seeming and sounding like bizarre and confusing versions of vacation rental listings, the results generated by Schmidt's AI are far less credible than the faces on Wang's website. (One included an image of a Dali-esque dining table, another incorporated the line "Minutes from Woods," and there is a garden or summer or a relaxing glow of all electricity products. ")
However, Schmidt also expects sites like his to make people question what they see online.
"Maybe we should all think in a few seconds before we assume something is real," he said.
